Stochastic Optimization for High-Dimensional Tracking in Dense Range Maps

M. Bray, E. Koller-Meier, P. Müller, N. N. Schraudolph, and L. Van Gool. Stochastic Optimization for High-Dimensional Tracking in Dense Range Maps. IEE Proceedings Vision, Image & Signal Processing, 152(4):501–512, 2005.
Earlier version

Download

3.6MB	840.5kB	3.5MB

Abstract

The main challenge of tracking articulated structures like hands is their many degrees of freedom (DOFs). A realistic 3-D model of the human hand has at least 26 DOFs. The arsenal of tracking approaches that can track such structures fast and reliably is still very small. This paper proposes a tracker based on stochastic meta-descent (SMD) for optimisations in such high-dimensional state spaces. This new algorithm is based on a gradient descent approach with adaptive and parameter-specific step sizes. The SMD tracker facilitates the integration of constraints, and combined with a stochastic sampling technique, can get out of spurious local minima. Furthermore, the integration of a deformable hand model based on linear blend skinning and anthropometrical measurements reinforces the robustness of the tracker. Experiments show the efficiency of the SMD algorithm in comparison with common optimisation methods.

BibTeX Entry

@article{BraKolMueSchetal05,
     author = {Matthieu Bray and Esther Koller-Meier and Pascal M\"uller
               and Nicol N. Schraudolph and Luc Van~Gool},
      title = {\href{http://nic.schraudolph.org/pubs/BraKolMueSchetal05.pdf}{
               Stochastic Optimization for High-Dimensional Tracking
               in Dense Range Maps}},
      pages = {501--512},
    journal = {IEE Proceedings Vision, Image \& Signal Processing},
     volume =  152,
     number =  4,
       year =  2005,
   b2h_type = {Journal Papers},
  b2h_topic = {>Stochastic Meta-Descent, Computer Vision},
   b2h_note = {<a href="b2hd-BraKolMueVanetal04.html">Earlier version</a>},
   abstract = {
    The main challenge of tracking articulated structures like hands is
    their many degrees of freedom (DOFs). A realistic 3-D model of the
    human hand has at least 26 DOFs. The arsenal of tracking approaches
    that can track such structures fast and reliably is still very
    small. This paper proposes a tracker based on stochastic meta-descent
    (SMD) for optimisations in such high-dimensional state spaces. This
    new algorithm is based on a gradient descent approach with adaptive
    and parameter-specific step sizes. The SMD tracker facilitates the
    integration of constraints, and combined with a stochastic sampling
    technique, can get out of spurious local minima. Furthermore,
    the integration of a deformable hand model based on linear blend
    skinning and anthropometrical measurements reinforces the robustness
    of the tracker.  Experiments show the efficiency of the SMD algorithm
    in comparison with common optimisation methods.
}}